Motivation

Asked to give workshop Deutsches Archäologisches Institut


Many (not all) of the archaeological teaching materials are focused on specific applications and University course materials

(nice comprehensive list: The didactic map of computational archaeology)


Something I’ve wanted to do for a while!

Inspiration

Good enough practices in scientific computing (Wilson et al. 2017)

  • data management
  • project organisation
  • collaboration (project portability)

R for Data Science textbook

Data Carpentry - R for Social Scientists

Teaching philosophy

Not meant to produce programmers and statisticians


Meant to enable researchers to confidently and reproducibly do their work in R


Programming concepts sparsely sprinkled throughout (as needed)

Teaching philosophy

Code along

  • can’t learn a programming language without doing

“Time to first plot” - Mine Cetinkaya-Rundel

  • early ‘wins’
  • keep learners motivated

Frequent formative assessments (a.k.a execises)

  • helps manage the pace
  • feedback on comprehension (workshops)

Teaching philosophy

tidyverse ecosystem mixed with some base R

  • gentlest entry to R (opinion!)
  • best-suited for data visualisations and analyses (fact-ish…)
  • best illustrations (fact!)

Teaching philosophy

tidyverse ecosystem mixed with some base R

  • gentlest entry to R (opinion!)
  • best-suited for data visualisations and analyses (fact-ish…)
  • best illustrations (fact!)

A cartoon Delorean, with several fuzzy monsters dressed in lab coats pouring date-times into the flux capacitor, with one holding a lubridate cheatsheet. One fuzzy monster is flying on a hoverboard, dressed like Marty McFly from Back to the Future. Title text reads 'lubridate: time control!'' Learn more about lubridate.

Artwork by @allison_horst.

Teaching philosophy

EDA > statistical tests

Teach statistical concepts, not tests (but also tests…)

  • general linear models
  • p-values…
    • statistical significance
    • NHST
    • continuous metric
  • confidence intervals
  • effect size
  • point/parameter estimates
  • practical implications (actual significance)

Teaching philosophy

Live coding during workshops

  • mirror learners’ environment (RStudio with no customisation)
  • what, how, why, repeat.
  • explain errors (both intentional and unintentional)

Notes from workshops are hosted on GitHub and

generated during the workshop with gitautopush

Central repo with new branch for each workshop

Modularity

Make the materials as flexible and customisable as possible:

  • duration of course/workshop
  • target audience of workshop (beginners, intermediate, advanced)
  • field of research (bioarchaeology, neolithic, paleobotany, etc.)

Modularity

Current modules

Basics

  • R basics
  • example workflow
  • Project organisation
  • Data cleaning

Exploratory Data Analysis (EDA)

  • Visualising data
  • Transforming data

Communication

  • Quarto document

Modularity

D cluster0 mat_basic R basics mat_workflow Ex workflow mat_basic->mat_workflow mat_project Project org mat_basic->mat_project mat_clean Cleaning data mat_project->mat_clean mat_viz Visualising data mat_clean->mat_viz mat_comm Communicating results mat_clean->mat_comm mat_transform Transforming data mat_viz->mat_transform mat_model Modelling data mat_viz->mat_model mat_transform->mat_comm mat_model->mat_comm

Modularity

Beginners

Brief workshop (ca. 2 hours)

  • R basics + Example workflow

1-day workshop (max. 6 hours)

  • Basics

2-day workshop

  • Basics + EDA
  • assignment 1

4-day workshop

  • Basics + EDA + modelling + communicating
  • assignment 1 + 2

Advanced users

1-day workshop (max. 6 hours)

  • cleaning + modelling (or topic-specific)

2-day workshop

  • cleaning + modelling + topic-specific
  • assignment 1 + 2

Materials

Slides

  • brief context for R(Studio) and Quarto
  • statistical concepts

Code-along materials

  • step-by-step code with explanations
  • used by the instructor while teaching
  • can also be used for self-study

Materials

sheep-data.csv

Sheep astraguli data Mediterranean Iron Age.

The contribution of Mediterranean connectivity to morphological variability in Iron Age sheep of the Eastern Mediterranean Sierra A. Harding, Angelos Hadjikoumis, Shyama Vermeersch, Roee Shafir, Nimrod Marom. bioRxiv 2022.12.24.521859; https://doi.org/10.1101/2022.12.24.521859

Accessed from: nmar79. (2023). nmar79/Med_Sheep_Astragals: v0.1 (v0.1). Zenodo. https://doi.org/10.5281/zenodo.10276147 (sheep_specinfo_20230824.csv)

Modifications: removes missing values and variables that can be calculated from existing variables

Materials

mortuary-data.xlsx

Burial data from northeastern Taiwan ranging from the Iron Age through the European colonization period.

Li-Ying Wang & Ben Marwick, (2021). Compendium of R code and data for “A Bayesian networks approach to infer social changes from burials in northeastern Taiwan during the European colonization period”. Online at https://osf.io/xga6n/

Accessed from: https://osf.io/zem9p (Kiwulan_Burials.xlsx - burials sheet)

Modifications: removes some variables that need cleaning (reduce cleaning complexity)

Assignments

Assignment 1: Case Study

Finding, importing, cleaning, and exploring/analysing a dataset.

1.1 Finding, importing, and cleaning

1.2 Exploring

1.3a Modelling (if modelling module is taught)

1.3b Communicating (if communicating module is taught)

Assignments

Assignment 2: Peer feedback

Can someone else reproduce your analysis?

Participants are paired up and need to reproduce each other’s work

Each participant produces a CODECHECK-style report

  • problems encountered
  • steps to solve problems

Feedback is incorporated into own project

If Git/GitHub is taught, this will be done via Git

Technical details

Site is built with Quarto

R environment captured with the renv R package

Hosted on GitHub Pages

Source code at github.com/rchaeology/RchaeoStats

F - Materials archived on Zenodo 10.5281/zenodo.13983363

A - Accessible online at rchaeology.github.io/RchaeoStats

I - Quarto and R (dependencies captured with renv)

R - Open, permissive licenses for materials and data

Roadmap

Development of additional modules

EDA

  • correlations

Modelling Data (in progress)

Communication

  • Quarto manuscripts (with extensions), presentations
  • Quarto parametrised reports

Research-specific modules

  • age-at-death and sex estimations
  • tidy dental data

Version control and collaboration

  • Git and GitHub


Better integration between the statistics and coding

To tidymodels or not to tidymodels?

What now?

Need community to contribute topic-specific modules (other contributions are welcome)

  • dendrochronology
  • radiocarbon dating (and others)
  • more

More iterations to improve existing materials

Acknowledgements

Also thanks to early test subjects at Österreichische Archäologische Institut

Thank you!

Join us at UnArchaeology.nl 7th (and 8th) November in Leiden!